Using Technology to Train Weather Forecasters
During the last decade, a revolution has occurred in weather forecasting. New satellites, Doppler radars, and other observing tools allow forecasters to view weather phenomena in ways and at a level of detail previously unimagined. Sophisticated computer systems integrate these observations with model predictions so that forecasters can analyze the information and display it in a multitude of combinations. To use this data effectively requires new interpretation skills based on an up-to-date knowledge of meteorology, particularly at geographic scales much smaller than forecasters have previously dealt with. However, advances in the science occur almost daily, and forecasters must frequently adjust their understanding of how weather works.
At the beginning of its modernization process, the National Weather Service (NWS) recognized the need for a comprehensive professional development program. Federal budget reductions also required that the training program be as cost-effective as possible. In 1988, the NWS turned for help to the University Corporation for Atmospheric Research (UCAR), this country's premier research facility in the atmospheric sciences. A year later, the NWS parent organization, the National Oceanic and Atmospheric Administration, and UCAR established the Cooperative Program for Operational Meteorology, Education and Training (COMET) to help update the scientific knowledge of its forecasters. Additional support for the COMET Program comes from the U.S. Air Force and the Navy.
The COMET Program's mission covers a broad spectrum of activities beyond professional development for the sponsors' forecasters (1,000 NWS forecasters and 5,000 military forecasters). Rather than describe all of these, this paper focuses on three areas in which the COMET Program has used technology over the last 10 years to create a comprehensive education and training environment to meet weather forecasters' needs. These three areas are our Residence Program, Distance Learning Program, and a Web-based facility that provides support for NWS trainers.
The COMET Residence Program
To ensure that its training is tailored for local conditions, the NWS has created a training position (called the Science Operations Officer or SOO) in each of the 121 local weather forecast offices. In addition to other duties, including producing forecasts themselves, the SOOs develop training plans for their staff, create and deliver local training, and adapt other training materials for their particular office's needs. Because the SOOs are critical to the NWS professional development efforts, the COMET Residence Program has concentrated mainly on training these trainers, as well as others from local NWS offices and those in the military who have specific training responsibilities.
One of the unique features of the Residence Program is that it has no regular in-house faculty. Courses are designed and coordinated by COMET staff, in consultation with lead instructors (typically a university professor and a SOO). Most of the course instructors are experts from research institutions, universities and the NWS.
Some of the visiting instructors lecture for only an hour or two, while others teach several days. By virtue of their expertise, these people are busy, and flying to Boulder for a single lecture can be a major interruption in their regular work. About five years ago we began using video teleconferencing to help experts who could not travel here to participate in the classes. A PictureTel Concorde 4500 system with a dedicated T1 line projects the instructor's image from t he classroom video projector onto a large screen. The remote instructor can see and hear the class on his or her own PictureTel system. Images and graphics are displayed by using a document camera, focusing the camera on a whiteboard, or showing electronic presentation files stored on a local PC. We annually host between 25 and 30 weeks of classes, and one or two instructors typically participate via teleconference each week.
A traditional lecture format (whether given remotely or in the classroom) makes up about half of a typical class day. In-class instructors have access to the usual array of instructional equipment: whiteboard, overhead projector, TV and videocassette recorder, slide projector and a fixed mount video projector that accepts input from the various instructor workstations. Students use wireless handheld microphones, and the instructors use lavaliere microphones.
The rest of the class day usually consists of collaborative problem-based learning exercises that examine cases relevant to the lecture materials. For these exercises, the students use nine Hewlett Packard C3000 workstations (for a class of 18 students). Each of these workstations has 100 Mbps access to our network and the Internet, and is paired with a Pentium PC for easy access to the Web and COMET distance learning modules.
The COMET staff has developed an extensive library of computer-based case studies that include integrated surface, upper air, satellite and radar data. A typical Residence Program course uses two or three case studies a week to support lecture topics. As interesting weather events occur, new case studies are created using COMET-developed software that can process standard weather data, as well as experimental data such as field study observations. We now have over 100 prepared case studies, many of which forecasters and university professors can access on the Web (www.comet.ucar.edu/resources/cases/index.htm). In addition, we also have an extensive archive of weather data that is used to prepare additional case studies as needed for both the residence courses and our other educational products.
Using these cases, students work with one or two partners to solve the "forecast problem of the day" and answer questions related to the lecture materials. Most of these exercises are studies that occur in what we call "displaced real-time," meaning that the problem follows the weather's evolution over a specific time period, and the students cannot access data for later time periods until they have answered the questions appropriate for the earlier analyses. Occasionally, an instructor will use an event unfolding in real-time (such as a hurricane making landfall) for the exercise. In both the real-time cases and the displaced real-time cases, the students access many of the same data sets and display software that they use at home in their forecast offices.
The classroom has undergone two hardware upgrades during its 10 years. The first moved the classroom from PC-based, proprietary hardware for ingesting and displaying data to a more open system of UNIX workstations. That upgrade cost $350,000 five years ago. The second upgrade occurred in 1999 when we purchased new PCs, a server, a hard disk array, and workstations to improve performance in running some of the more complicated software packages. This last upgrade cost about $350,000, as well. While expensive, these recurring upgrades are necessary to ensure that our students use systems and software that are similar to what they have at home or will have in the near future. Planning for upgrades has become an essential part of our budgeting process.
In addition to learning the importance of keeping our facilities technologically current, we have also learned that simple technologies can be used to make our students' lives easier and their education more productive. Because they are trainers themselves, Residence Program students are expected to tailor the knowledge and skills they have gained from the courses for their own forecasters. However, the SOO's jobs include many duties besides training, and they simply do not have time to convert the boxes full of course materials into presentations or self-study modules.
Two years ago, the SOOs requested the class materials in electronic form. The difficulty was that we had no control over the formats our volunteer instructors used for their presentations. In the past, the lecture materials usually consisted of stacks of viewgraphs (often hand-written the day before the class) that we would copy for the students. During the last two years, we have strongly encouraged the instructors to convert their presentation materials to electronic format, helping them when necessary. We then link the presentations to a course Web page that students and others can access (http://www.comet.ucar.edu/class/index.html). We have found that, once the instructors are coached in developing their first lesson, they rapidly adopt presentation technologies for other classes.
The Distance Learning Program
Sending thousands of forecasters to residence courses would be prohibitively expensive, so the COMET Program also develops distance learning materials for use by forecasters at their duty stations. Efforts to date have focused largely on creating interactive, multimedia computer-based learning (CBL) modules, but we are continually experimenting with other technologies.
A typical CBL module contains 4 to 8 hours of highly interactive instruction and uses a mix of case studies, graphics, animations, audio and video. Text and spoken dialogue introduce concepts that are reinforced by graphic materials such as time-sequenced satellite and radar data, and videos demonstrating laboratory experiments or showing experts explaining concepts. At various points throughout each module, students can practice using the concepts covered by answering questions and working through realistic case studies. In many cases, additional material is presented in the module, often by an expert in the field, when the student requests more detailed information or incorrectly answers a question. This instructional design follows the principles of cognitive apprenticeship and situated cognition described by Collins (1991) and Brown, Collins and Duguid (1989).
We produced our first CBL modules on laser disk. When reliable digital video and audio became feasible about six years ago, we changed to CD-ROM as the delivery medium. Now that streaming technologies have become more reliable on the Web, we develop more of our modules for that environment (http://meted.ucar.edu/modules.htm) so we can more easily disseminate, update, and correct materials. The CD and Web-based formats are much more convenient for forecasters, who can access the modules at their forecasting workstations during slow periods or even at home. However, some of the forecast offices still have slow Internet connections, and Navy forecasters on ships cannot access the Web for extended periods. To solve this problem, we periodically compile materials produced recently and distribute them on CDs.
Developing a CBL module is a complex process, requiring instructional designers, meteorologists, hydrologists, graphics and media specialists, computer scientists, and other experts in the specific field addressed by each module. Developing one 4-6 hour module takes approximately one year and costs about $250,000. At the end of 1999, we had produced a total of 37 modules on various topics including Doppler radar interpretation, marine forecasting, fire weather forecasting, numerical weather prediction, etc.
In addition to the modules, we have experimented recently with teletraining and Webcasts as part of our distance education program. We have conducted the teletraining lessons using a software package called Telewriter 2000 (from Optel Communications, Inc.). The instructorcan connect through the Telewriter software with an unlimited number of offices ó we have worked with as few as three and as many as eight. Each office participating in the lesson needs a copy of the same software, the lesson files (downloaded before class), and a hardcopy of any additional materials provided (complicated graphics, survey forms, etc.).
Forecasters gather around a workstation where they can view the presentation and listen to the instructor, who usually lectures briefly and then guides the students as they work on several problems. Both the instructor and students can point out features on graphics, play and discuss animations, and listen to sound files. The main advantage of this delivery method is that a considerable number of forecasters can receive highly interactive training on short, focused subjects over a relatively short time frame.
Webcasts are recorded audio lectures that are coordinated with electronic presentations and, as the name implies, are delivered over the Web. As the audio file with the instructor's lecture plays, the corresponding text and graphic material are displayed. This technology allows us to record and archive residence course lectures and then put the materials on the Web as we find time. We usually have the instructor record the presentation separately from the in-class lecture to provide better sound quality.
Ten years of producing modules have taught us a number of lessons. As is the case with our Residence Program, we have found that our Distance Learning Program must constantly prepare for technological changes. When we began producing modules, laser disks were the only practical way to include video, audio, and data graphics in the lessons, but they also had significant disadvantages. One of these was the cost; producing modules on laser disk was 10 times more expensive than pressing CDs is now. Had we not kept up with the evolution in CD technology, we would not have been ready to change to these improved delivery modes. Our efforts to prepare for new technologies well in advance have allowed us to undertake risk reduction measures, preventing us from making major mistakes. Adopting new technologies always presents unexpected difficulties. The key is to minimize problems as far in advance as possible.
We have also learned that we can make our production process more efficient by reducing our dependence on subject matter experts. These experts are busy people and the production process could come to a halt for months until they have time to devote to the module development. Additionally, subject matter experts often know almost too much. What was supposed to be a 4-hour module can rapidly turn into a 14-hour module. We have found our process is improved by having a knowledgeable staff who can help outline the module and fill in the broad content, while leaving the finer points to the subject matter experts.
We have not found answers to all of our challenges. We still struggle with how to develop products that meet the needs of all our diverse users. Surveys of our target audience help refine our products, but we cannot please everyone. For example, several years ago we received feedback that our modules were much too long. Forecasters typically do training during slack periods that rarely last longer than an hour and do not occur every day. A module that is 10 hours long can literally take a month or more to complete. Finding that balance between conveying too much information and not enough is difficult, especially since our audience has such a wide range of background knowledge and skills. For example, few of the military forecasters have a degree in meteorology, while most of the NWS employees do. We are constantly challenged to find the right mix of "how-to" information and scientific background, mathematical foundations versus non-mathematical concepts, longer and more detailed content versus shorter and less rigorous treatments. In the future we plan to try providing different paths through modules for forecasters with different backgrounds. Multiple paths will undoubtedly complicate the dev elopment process but will hopefully serve our sponsors' needs better.
Over the last year, the COMET Program has also experimented with new ways to support the SOOs in their training efforts at their local offices. Few of them have prior experience in training, and they are given minimal preparation when they are hired as SOOs. Because they are geographically separated, they have little contact with each other.
To help with this problem, we have developed a Web-based SOO Training Resource Center (TRC). The goal of the TRC is to enable the SOOs to share training materials they have developed and other resources they have found useful for meeting their training responsibilities. In addition, we hope to provide learning opportunities for those who are unfamiliar with educational theory, instructional design, and other components of training. Finally, we also believe that the TRC can foster a greater sense of community in which the SOOs share ideas and the more experienced members mentor newer members.
The TRC consists of the entry page (http://meted.ucar.edu/resource/soo/index.htm) that describes the other three main pages: Catalogue, Forum, and Other Resources. The Web-based catalogue has proven to be very successful. Similar attempts to create paper-based catalogues several years ago failed, but the online catalogue apparently provides the immediacy and simplicity that the SOOs require. Prior to the TRC, a search of NWS regional Web sites identified only five SOO-developed training materials that other SOOs could access, assuming they would think to search sites outside their own region. Eight months after the TRC Web site opened, the Catalogue page had 63 entries posted by 26 different SOOs. During the same period, 19 of 36 SOOs responding to a survey indicated that they had used the Catalogue to contact another SOO to obtain a copy of a training product.
The Other Resources page contains a variety of resources, including copies of training plans developed by SOOs who are considered by their peers to be good trainers, links to resources about developing presentations, links to instructional design materials, and teletraining schedules. The Forum page is intended as a place for holding discussions that we feel are key to building a community of practice among the SOOs. Four threaded discussions have been held so far. The first two topics discussed ideas or techniques SOOs had found most effective in their training programs. The last two topics were more instructional; one was related to instructional design and the other dealt with developing and delivering teletraining.
Overall, the TRC is a good example of an idea that had to wait for the right technology. The previous attempts to establish a paper-based catalogue of training materials failed because the SOOs disliked filling out the overly complicated submission forms, and because updating the paper copies took too long for the catalogues to be useful. The immediacy of the Web solved both of these problems.
The last 10 years have been an interesting time of experimentation for the COMET Program. If we have learned one thing, it is that it is not easy to provide training for such a diverse workforce in a rapidly evolving field. Training also is more complicated now compared to a few years ago because the technological options are so numerous and are constantly changing. Being flexible is critical. Training organizations must be ready to adapt to new technologies, new requirements from sponsors, and new understandings of their audiences' needs. Ultimately, what works best comes back to good instructional systems design: know your audience, understand their needs and limitations, and use the most suitable technology and great instructional strategies to create an interactive training environment to meet those needs.
Dr. Victoria Johnson is the Outreach Program Manager for the Cooperative Program for Operational Meteorology, Education and Training (COMET) at the University Corporation for Atmospheric Research (UCAR). She has an M.S. degree from the University of Wyoming in Atmospheric Science and an Ed.D. from Nova Southeastern University in Instructional Technology and Distance Education.
Brown, J.S., Collins, A., and Duguid, P. 1989. Situated Cognition and The Culture of Learning. Educational Researcher, 18(1), 32-42.
Collins, A. 1991. Cognitive Apprenticeship and Instructional Technology. In L. Idol & B. F. Jones (Eds.), Educational Values and Cognitive Instruction: Implications for Reform (pp. 453-494). Hillsdale, NJ: Lawrence Erlbaum Associates.
This article originally appeared in the 06/01/2000 issue of THE Journal.